Demand model recalibrations
Nowcast
Last week we estimated a 88 Bcf injection for the storage week ending 16 October, against the EIA's reported 77 Bcf. After the report we reviewed the data behind our estimate and found a way to improve how our Electric Power nowcast uses the EIA's grid data. That improvement is in today's release. Our forecast models are unchanged.
The update makes better use of the latest hourly grid data (EIA-930) and assigns gas-fired generation to regions more precisely. As a result, our daily Electric Power estimates settle faster and are more stable from one run to the next, especially in the South Central region.
| Storage week ending | EIA Reported (Bcf) | Previous method (Bcf) | Updated Method (Bcf) |
| 16th October | 77 | 87 | 85 |
| 23rd October | not reported yet | 76 | 73 |
Rerun on today's data, the update lowers last week's estimate from 87 to 85 Bcf (later data had already moved our published 88 to 87). That is still above the EIA's 77 Bcf, and we are continuing to review the rest of last week's difference. For the current week, modelled Electric Power burn rises by 0.32 Bcf/d, mostly in the South Central region, and our storage estimate falls from 76 Bcf to 73 Bcf as of 26 October.
The change is one of timing within the month, not of level. Monthly Electric Power demand moves by less than 0.2 Bcf/d: August 48.98 → 48.96, September 43.12 → 42.94 and October to date 34.06 → 33.99 Bcf/d. Months the EIA has already published stay aligned to its figures. Residential, Commercial and Industrial demand are unchanged.
